A wide variety of different techniques for testing software code are known. These conventional code testing techniques have significant drawbacks, particularly when utilized to test application code for applications executed in virtual machines, containers and other compute resources of cloud infrastructure. For example, the application code is typically tested using prefabricated or otherwise static test data. Such an arrangement is problematic in that various bugs, weaknesses or other types of failures in the application code may not be exposed until after the application code is deployed and executed in the cloud infrastructure using dynamically-varying data. Moreover, additional difficulties can arise when testing upgraded application code. For example, conventional techniques for testing upgraded application code in the context of live upgrades to applications running in cloud infrastructure can be unduly disruptive.